security<\/p>\n
Corma CEO tells The Reg it’s building ‘One ring to rule them all, for the defenders to have this power’<\/p>\n
Corma CEO Alon Pluda says his AI security startup aims to close the “defense gap,” where models are better at offensive security. He tells the story of one customer, a security executive who was walking his dog when he received a notification on his watch from a Corma agent.<\/p>\n
\u201cIt said, ‘I just caught a live attack. I need your permission to block it,’\u201d Pluda told\u00a0The Register in an interview.\u00a0<\/p>\n
The security boss approved the agent\u2019s action; the agent blocked the malware and the attacker from moving across the company\u2019s network and mitigated the intrusion in under 10 minutes, Pluda said. <\/p>\n
The customer later described “walking outside with his dog, and blocking a real-live attack with his AI coworker” as “one of the most magical moments of his year,” Pluda recalled.<\/p>\n
Pluda founded Corma about a year ago. And yes, all you Lord of the Rings nerds, the company gets its name from the Elven word for \u201cring.\u201d \u201cWe\u2019re building the one ring to rule them all, but this time for the defenders to have this power.\u201d<\/p>\n
Earlier this week, the company announced $60 million in seed funding led by Sequoia Capital, alongside Khosla Ventures and Coatue. He told us that his startup is working with Fortune 100 companies, and training models to achieve \u201csuperintelligence for defensive cybersecurity.\u201d<\/p>\n
Models from OpenAI, Anthropic, and Google are \u201camazingly good\u201d at coding and language, and this includes finding and fixing bugs, and orchestrating tools across multi-step workflows, he explained.\u00a0<\/p>\n
\u201cWhen you combine it with agentic capabilities, they move from being incredible vulnerability researchers to end-to-end attackers<\/a>,\u201d Pluda said. \u201cSo inherently, what we\u2019ve seen in the last few months is the models getting exponentially better at offensive security, like we saw with the OpenAI and Hugging Face incident<\/a>.\u201d\u00a0<\/p>\n But these same models aren\u2019t as skilled at carrying out defensive security tasks that don\u2019t involve scanning code for vulnerabilities and misconfigurations, he said. \u201cThe vast majority of defensive security tasks don\u2019t have anything to do with code.\u201d<\/p>\n Corma recently tested four frontier models – Claude Opus 4.8, GPT-5.5, Grok 4.3, and DeepSeek V4 – as both attackers and defenders across the same fake company and its networks, built to closely mirror a multi-business enterprise. The attacker\u2019s task was to plant a backdoor and the defender\u2019s task was to find it and stop the attack.\u00a0<\/p>\n Closing the defensive gap<\/p>\n Corma ran all four models against each other in every attacker and defender pairing, including each model against itself, with 15 independent engagements per pairing for 241 scored engagements. Across all of these, the models successfully implanted a persistent backdoor in 85 percent of their runs. However, these same models only detected 19 percent of attacks.<\/p>\n \u201cThat speaks to the inherent imbalance we are trying to solve,\u201d Pluda said. \u201cThe general foundation models are getting exponentially better at offensive security, but haven’t been able to improve on the same rate on defensive security. So our mission is to close this gap, and make sure the defenders win in this intelligence-versus-intelligence game – or war.\u201d\u00a0<\/p>\n Corma calls this the defensive gap, and says it has to do with the data these models are trained on and the objectives they are trained against, which lend themselves to offensive security.<\/p>\n